1 City

2 County

3 Should You Become a Data Scientist?

There are tons of jobs, high Salary and much growth. IBM Predicts Demand For Data Scientists Will Soar 28% By 2020. [3]. Let’s check the facts:

4 Employment Projections

# add geom_hline showing "Total media wages"
# # Change the "horizontal justification" to be left-justified with hjust 

mycols <- c("blue", "red")
ggplot(data=occupation, mapping=aes(x=industry, y=median_wages17, fill=compu)) +
  stat_summary(fun.y=mean, geom="bar") +
  scale_y_continuous(labels=dollar) +
  labs(title="Median Wages by Industry in 2017", x="", y="") + 
  coord_flip() + 
  theme(legend.position="none") + 
  scale_fill_manual(values=mycols) + 
  geom_text(aes(label = median_wages17), color = "black", hjust=-0.1) +
  geom_hline(yintercept = 37690, color="red", size=1, linetype="dashed")

The median annual wage for computer and information technology occupations was $84,580 in May 2017, which was more than double the median annual wage for all occupations of $37,690.[4]

# remove all occupations

occupation <- occupation[2:23,]

occupation$industry <- fct_reorder(occupation$industry, occupation$change2026, mean, na.rm=TRUE)
ggplot(data=occupation, mapping=aes(x=industry, y=change2026, fill=compu)) +
  stat_summary(fun.y=mean, geom="bar") +
  scale_y_continuous() +
  labs(title="Employment Jobs Growth Projection by Industry 2016-2026 ", x="", y="") + 
  coord_flip() + 
  theme(legend.position="none") + 
  scale_fill_manual(values=mycols) + 
  geom_text(aes(label = change2026), color = "black", hjust=-0.1) 

  # geom_hline(yintercept = 37690, color="red", size=1, linetype="dashed")

Employment of computer and information technology occupations is projected to grow 13 percent from 2016 to 2026 in the US, faster than the average for all occupations.

occupation$industry <- fct_reorder(occupation$industry, occupation$perchange206, mean, na.rm=TRUE)

ggplot(data=occupation, mapping=aes(x=industry, y=perchange206, fill=compu)) +
  stat_summary(fun.y=mean, geom="bar") +
  scale_y_continuous() +
  labs(title="% Employment Growth Projection by Industry 2016-2026 ", x="", y="") + 
  coord_flip() +
  theme(legend.position="none") +
  scale_fill_manual(values=mycols) +
  geom_text(aes(label = perchange206), color = "black", hjust=-0.1)

rm(mycols)

These occupations are projected to add about 607k new jobs.

5 Dinamic Visualizations

6 Dot Density Map

# Tech
# Set zero margins in plot window, so that map fits
par(mar=c(0,0,0,0))
# plot(zips.shp, lwd=0.1)

# One dot per 10 people
plotvar <- df.17$tech/10

# Calculated random dots for each zip code and store the coordinates in cadots.rand
cadots.rand <- dotsInPolys(zips.shp, as.integer(plotvar), f="random")

# Plot a specific region - zooms in LA conty
plot(zips.shp, lwd=0.1, xlim=c(-118.510000,-117.820000), ylim=c(33.710000,34.860000))
plot(cadots.rand, add=T, pch=19, cex=0.1, col="#000080", border="red") # Blue ---- "#09900d" Green

# Info
# Set zero margins in plot window, so that map fits
# par(mar=c(0,0,0,0))
# # plot(zips.shp, lwd=0.1)
# 
# # One dot per 10 people
# plotvar1 <- df.17$info/10
# 
# # Calculated random dots for each zip code and store the coordinates in cadots.rand
# cadots.rand <- dotsInPolys(zips.shp, as.integer(plotvar1), f="random")
# 
# # Plot a specific region - zooms in LA conty
# plot(zips.shp, lwd=0.1, xlim=c(-118.510000,-117.820000), ylim=c(33.710000,34.860000))
# plot(cadots.rand, add=T, pch=19, cex=0.1, col="#8B0000", border="red") # Blue ---- "#09900d" Green # #8B0000 Red # #FF4500 Orange

# Professional 
# Set zero margins in plot window, so that map fits
# par(mar=c(0,0,0,0))
# # plot(zips.shp, lwd=0.1)
# 
# # One dot per 10 people
# plotvar2 <- df.17$prof/10
# 
# # Calculated random dots for each zip code and store the coordinates in cadots.rand
# cadots.rand <- dotsInPolys(zips.shp, as.integer(plotvar2), f="random")
# 
# # Plot a specific region - zooms in LA conty
# plot(zips.shp, lwd=0.1, xlim=c(-118.510000,-117.820000), ylim=c(33.710000,34.860000))
# plot(cadots.rand, add=T, pch=19, cex=0.1, col="#FF4500", border="red") # Blue ---- "#09900d" Green # #8B0000 Red # #FF4500 Orange

rm(plotvar)
rm(cadots.rand)
rm(cacounties)
rm(cacounties.shp)
rm(cnty.zips)
rm(cnty.zips.17)

Shows a high density of tech jobs in West LA, West Hollywood, Burbank, Westlake and Pasadena areas in 2017 [6] [7] [8]
Each dot = 10 tech jobs

# install imageMagick
# https://www.imagemagick.org/script/download.php#windows

# Tutorials
# https://ryanpeek.github.io/2016-10-19-animated-gif_maps_in_R/
# https://cran.r-project.org/web/packages/magick/vignettes/intro.html

# library(magick) # this is call to animate/read pngs

# save all png files to path folder

# create gif
# list.files(path = "C:/Users/joao_/Downloads/Gif/", pattern = "*.png", full.names = T) %>%
#   map(image_read) %>% # reads each path file
#   image_join() %>% # joins image
#   image_animate(fps=100) %>% # animates, can opt for number of loops
#   image_write("tech_100.gif") # write to current dir

7 Interactive Zip Code Map

8 Correlation Between Education and Income

# Adds Title / Lables
p5 = p4+
scale_x_continuous(name = "City Human Capital Index (CHCI)")+
scale_y_continuous(name = "Log Income, 2016") +
ggtitle("CHCI and Median Income Per Zip Code")
p5

## Warning: Removed 7 rows containing non-finite values (stat_smooth).

## Warning: Removed 7 rows containing missing values (geom_point).

## Warning: Removed 7 rows containing missing values (geom_point).

## Warning: Removed 7 rows containing missing values (geom_text_repel).

Association between the 2012 CHCI [9] and the 2016 median household income [10] of 282 zip codes in L.A. County. We can see a very strong correlation correlation between the human capital level [23] and the household income for the zip code in L.A. It demonstrates the importance to invest in education (chci) in order to create more productive workforce and more high paying jobs.

################

## Need to clean code bellow

# str(df.17)
# df.17$GEOID <- as.factor(df.17$GEOID)
# names(masterzip) <- c("GEOID", "name")
# z <- inner_join(df.17, masterzip, by = "GEOID")
# str(z)
# top_zip <- z %>% arrange(desc(per)) %>% top_n(20, wt=per)
# 
# names(top_zip) <- c("zip", "V1",    "year",  "tech",  "info",  "prof",  "per",   "total", "name")
# str(cor)
# str(top_zip)
# top_zip$zip <- as.numeric(top_zip$zip)
# 
# cor <- inner_join(top_zip, cor, by="zip")
################
# NEW

# Create the scatter plot with regression line
# 
# p7 <- ggplot(cor, aes(x=chci, y=estimate)) +
#   geom_point(colour = "blue", size=4) +
#   labs(title = "Fig. 8 ",
#        x = "City Human Capital Index", y = "Median Household Income ($)") +
#   scale_x_continuous(limits = c(85, 175)) +
#     geom_smooth(colour = "red", method=lm,   # Add linear regression line
#               se=FALSE)    # Don't add shaded confidence region
# p7

# p1 <- ggplot(cor, aes(x=chci, y=log(estimate)))+
#   geom_point(colour = "red")
# p1
# p2 <- p1 +
#   geom_smooth(mapping = aes(linetype = "r2"),
#               method = "lm",
#               formula = y ~ x , se = FALSE,
#               color = "black")
# p2 + geom_point()
# p2 + geom_point(shape = 1, size = 2) # Changes to shape = 1
# p3=p2 + geom_point(shape = 1, size = 3, stroke=1.5)
# p3
# # Labels overlap - use ggrepel
# # install.packages("ggrepel") 
# library("ggrepel") # Repel overlapping text labels away from each other.
# p4 <- p3 + geom_text_repel(aes(label = zip), size=2) # Uses greppel text repel
# p4



# Adds Title / Lables
# p5 = p4+
# scale_x_continuous(name = "Capital Human City Index")+
# scale_y_continuous(name = "Log Income, 2016") +
# ggtitle("CHCI and Median Income Per Zip Code")
# p5
# 
# summary(cor)
# chci media 163

################
# str(df.17)
# df.17$GEOID <- as.factor(df.17$GEOID)
# names(masterzip) <- c("GEOID", "name")
# z <- inner_join(df.17, masterzip, by = "GEOID")
# str(z)
# top_zip <- z %>% arrange(desc(per)) %>% top_n(30, wt=per)
# str(cor)
# str(top_zip)


# # Creates vector with list of top tech zip codes
# top_zip$GEOID -> zip.label
# 
# # Only shows labels from list created
# 
# library(ggplot2)
# library(readxl) 
# library(tidyverse) # Includes ggplot2
# library(dplyr)
# 
# p6 = p5 + geom_text(aes(label = zip),
#             color = "gray20", data = filter(cor, cor$zip %in% zip.label))
# p6
# 
# 
# p4 = p3 + geom_text(aes(label = Country),
#             color = "gray20", data = filter(cpigdp, Country %in% country.label))

9 Conclusions

• This research shows that Data Science is a promissing field with tons of jobs, high salary and great growth.

How to Become a Data Scientist in 8 Easy Steps [11]

1. Get good at stats, math and machine learning
2. Learn to code
3. Understand databases
4. Master data munging, visualization and reporting
5. Level up with Big Data
6. Experience, practice and meet fellow data scientists
7. Internship, bootcamp or get a job
8. Follow and engage with the community

10 References

1. [1] PHOTO BY TYLER MERBLER - https://www.flickr.com/photos/37527185@N05/13934368189/
2. [2] PHOTO BY PAOLO GAMBA - https://www.flickr.com/photos/abukij/18012989253/
3. [3] IBM Predicts Demand For Data Scientists- https://www.forbes.com/sites/louiscolumbus/2017/05/13/ibm-predicts-demand-for-data-scientists-will-soar-28-by-2020/#6d337fec7e3b
4. [4] Employment Projections - Data from the Bureau of Labor Statistics - https://www.bls.gov/emp/tables/emp-by-major-occupational-group.htm
5. [5] Tech industry employment was calculated using the following: (1) Information jobs (NAICS: 51), and (2) Professional, Scientific, & Technical Skills (NAICS: 54). Data source: Quarterly Census of Employment and Wages (QCEW) developed through a cooperative program between the states and the U.S. Bureau of Labor Statistics. These data are summarized by Industry Sector (2-digit NAICS).
6. [6] Dowloaded shape files for zip code boundaries from the County Data Portal https://data.lacounty.gov/Geospatial/ZIP-Codes/65v5-jw9f
7. [7] Downloaded Zip Code List from - http://file.lacounty.gov/SDSInter/lac/1031552_MasterZipCodes.pdf
8. [8] Inspired by How to Make Dot Density Maps in R - Nathan Yau - https://flowingdata.com/2014/08/28/how-to-make-dot-density-maps-in-r/
9. [9] City Human Capital Index - http://www.anderson.ucla.edu/centers/ucla-anderson-forecast/projects-and-partnerships/city-human-capital-index
10. [10] Median income by zip code data from the American Community Survey obtained using the tidycensus R package. Request API at https://api.census.gov/data/key_signup.html
11. [11] How to become a data scientist in 8 easy steps. - https://insidebigdata.com/2014/11/14/become-data-scientist-8-easy-steps/

11 Acknowledgements

• This project was largely inspired by conversations with professor William Yu, economist at UCLA ANDERSON FORECAST
• Dynamic Visualizations were inspire by Google Public Data Explorer.

# install.packages("kableExtra")
# 
# library(kableExtra)
# 
# kable(df.17)
# 
# # The code below will automatically include the image in the rmarkdown document
# kable(dt, "latex", booktabs = T) %>%
# column_spec(1, bold = T) %>%
# kable_as_image()

rm(m)
rm(zips.shp)
rm(income)
rm(lacountytech)
rm(lacounty_data)
rm(popup)